Logo Lanfrica
  • Home
  • Atlas
  • Insights
  • Docs
  • Sign in

© 2026 Lanfrica. All rights reserved. All copyrights of the resources shown on the Lanfrica website belong to the original copyright holders, unless explicitly stated otherwise.

TinyML-Based Algerian Arabic Darija Keyword Spotting for Offline Smart-Home Control: Dataset, Model, and Hardware Deployment

Domain:

natural language processing

Record type:

dataset
Creator:
kai
Editor:
kai
Publisher:
Zenodo
Host:avatar
This dataset comprises 5,100 audio recordings (16 kHz, 16-bit, mono, WAV, 2.0 s duration) collected from 100 native Algerian Arabic Darija speakers across four indoor environments (living spaces, kitchens, entrance halls, utility rooms). It supports keyword spotting (KWS) for 16 compound smart-home voice commands spanning five domains (lighting, climate, water/gas/electricity mains, access control, television) plus a background noise class for rejection training. Each command follows a two-word action–object structure (e.g., chaal-dow "turn on light", eftah-gaz "open gas valve"), yielding four shared verb prefixes (chaal-, tafi-, eftah-, eghlak-) and eight shared object suffixes, introducing realistic acoustic overlap for dialectal morphological analysis. Total duration: 170 minutes (160 min commands + 10 min environmental noise).Split: 4,416 training / 384 test files (≈92%/8%, speaker-disjoint for commands).Balanced: 300 samples per class (3 repetitions × 100 speakers per command; 300 noise files). Label Algerian Darja English Meaning Home Domain Revised IPA chaal-dow شعل الضو turn on the light Lighting /ʃaʕːel dˤːaw/ tafi-daw طفي الضو turn off the light Lighting /tˤafːi dˤːaw/ chaal-clima شعل الكليما / الكليماتيزار turn on the air conditioning Climate /ʃaʕːel lklima/ tafi-clima طفي الكليما / الكليماتيزار turn off the air conditioning Climate /tˤafːi lklima/ afta7-elma افتح الما open the main water supply Water (main valve) /eftaħ lmaː/ ekfel-elma اقفل الما close the main water supply Water (main valve) /eqfel lmaː/ chaal-tv شعل التيلي turn on the television TV /ʃaʕːel tːili/ tafi-tv طفي التيلي turn off the television TV /tˤafːi tːili/ eftah-garage افتح الكاراج open the garage door Garage /eftaħ lɡaraːʒ/ eghlak-garage اغلق الكاراج close the garage door Garage /eɣlaq lɡaraːʒ/ eftah-lbab افتح الباب open the (main entry) door Door /eftaħ lbaːb/ eghlak-lbab اغلق الباب close the (main entry) door Door /eɣlaq lbaːb/ eftah-gaz افتح الغاز open the main gas supply Gas (main valve) /eftaħ lɡaːz/ eghlak-gaz اغلق الغاز close the main gas supply Gas (main valve) /eɣlaq lɡaːz/ tafi-disjoncteur طفي الديجونكتار cut the main electrical circuit breaker Electricity (main breaker) /tˤafːi dːiʒɔ̃ktœːr/ talaa-disjoncteur طلع الديجونكتار restore the main electrical circuit breaker Electricity (main breaker) /tˤalːaʕ dːiʒɔ̃ktœːr/ noise — background noise / no command Rejection class —

Visit

doi.org

Tasks

keywordsspeech processing

Languages

Arabic, Algerian SpokenTafi

Licenses

Creative Commons Attribution 4.0 Internationalhttps://creativecommons.org/licenses/by/4.0/legalcode

Similar

Twi Keyword Spotting Dataset for Iot ApplicationTinyML for smart agriculture: Comparative analysis of TinyML platforms and practical deployment for maize leaf disease identificationKaidtarek/Voice-Assisted-Smart-Home-With-TinyML-For-Low-ResourceA Deep Learning Framework for Arabic Continuous Speech Keyword Spotting in Low-Resource Settings Using Isolated-Word Keyword Spotting and Posterior Probability FunctionsEnhancing Accessibility of Feedback Collection in ML Models Through Keyword Spotting: A Moroccan Darija ModelHalimatou10/Fulfulde-Keyword-Spotting

Twi Keyword Spotting Dataset for Iot Application

A voice command dataset for a low resource language from Ghana , Twi.

TinyML for smart agriculture: Comparative analysis of TinyML platforms and practical deployment for maize leaf disease identification

International audience

Maize, a vital global food crop, is susceptible to var

Kaidtarek/Voice-Assisted-Smart-Home-With-TinyML-For-Low-Resource

Voice-Assisted | Smart-Home | TinyML | Low-Resource # TinyML Keyword Spotting for Algerian Darja

A Deep Learning Framework for Arabic Continuous Speech Keyword Spotting in Low-Resource Settings Using Isolated-Word Keyword Spotting and Posterior Probability Functions

Continuous Speech Keyword Spotting (CSKWS) presents a challenging paradigm shift from isolated-word

Enhancing Accessibility of Feedback Collection in ML Models Through Keyword Spotting: A Moroccan Darija Model

Halimatou10/Fulfulde-Keyword-Spotting

Keyword Spotting system for Fulfulde livestock vocabulary # KWS Fulfulde ## Description Ce projet